Use of a Sterile Collection Process to Reduce Contaminated Peripheral Blood Cultures

This initiative was conducted in our pediatric ED at a 280-bed tertiary, freestanding children’s hospital with ∼51 000 annual ED visits and 11 000 annual admissions per year. About 1700 peripheral blood cultures are obtained annually. Blood cultures in the baseline period were collected via a clean procedure by nursing staff, unless a failed attempt necessitated the involvement of an intravenous (IV) therapy team. The baseline period was from June 2, 2018, through December 31, 2018. The intervention period was from January 1, 2019, through April 30, 2020.

Stakeholders for this QI initiative included pediatric emergency medicine physicians, pediatric hospitalists, nursing staff, laboratory personnel, and hospital leadership. The intervention group included all children with peripheral blood cultures obtained in the ED.

A QI committee that included pediatric hospitalists, an emergency medicine physician, emergency medicine nursing leadership, and laboratory leadership was created. Key drivers are listed in Fig 1.

Nursing leadership, with input from nursing staff, created a sterile collection process using a blood collection kit. Sample kits were trialed in the ED, and changes were made on the basis of nursing feedback. The kit included a sterile IV start kit, sterile towels, and a sterile transfer device. Nurses were instructed to obtain sterile gloves, blood culture bottles, and other supplies separately. An audit card was included in the kit with specific instructions on the steps of the sterile process (Supplemental Fig 5). The audit card also included date and time of the blood culture specimen, age of the patient, whether point-of-care ultrasound was used for the IV start, whether the nurse auditing the procedure was the one performing the procedure, and feedback about the kit. The cost for each blood culture collection kit was $10.71.

A step-by-step video was created to educate nurses on the sterile collection technique. The video also reiterated the impact of contaminated blood cultures on patients and families. The video was uploaded into a portal accessible to nurses, and they were instructed to watch the video from January 17, 2019, to February 28, 2019. In addition, nurses also received one-on-one education about the collection process from February 4, 2019, to February 28, 2019. During the initial 6 months of implementation, QI team members attended nursing huddles and rounded in the ED several times per week to answer questions and elicit feedback about the new process.

For each blood culture that grew contaminants, we sent individualized feedback via e-mail to the nurse who obtained the culture. In the e-mail, we detailed the impact of the contaminated blood culture on the patient, including additional laboratory workup, procedures performed, antibiotics started, and return hospital visits. Individualized feedback was started on April 15, 2019.

Before implementation, we sent an anonymous survey via Survey Monkey from November 5, 2018, to December 31, 2018, to nurses in the ED. We sought to assess their knowledge regarding contamination rates in our ED, the impact of contaminated blood cultures in children, and obtain feedback on how to reduce contaminated blood cultures, including their opinion on using a sterile peripheral blood culture collection process.

We sent another survey 6 months after implementation to obtain feedback on the sterile collection process. On the basis of the survey results, we changed 1 step in the process from double gloving and removing the outer gloves after cleaning the venipuncture site to, instead, switching out sets of sterile gloves because nurses found it difficult to palpate veins while double gloving. An official announcement was sent out on November 11, 2019, with an updated video on November 21, 2019.

The QI committee reviewed nursing feedback obtained from audit cards monthly and addressed specific concerns listed with nursing staff.

Rates of contaminated peripheral blood cultures were plotted on a run chart and shared monthly with ED nursing staff.

Monthly blood culture contamination reports for baseline and intervention periods were obtained from the laboratory database. Information obtained from the laboratory report included patient name, medical record number, date and time of blood culture, unique accession number of culture, site of culture obtained (peripheral versus central), identity of the nurse who obtained the culture, if there was growth of a contaminant, and name of the contaminant organism. A contaminated peripheral blood culture was defined as growth of an organism considered as a commensal by our laboratory standards and confirmed by chart review (Supplemental Table 2). Growth of the same organism in >1 culture in the same patient was considered a true infection per laboratory standards. Cultures that were treated as true pathogens by the medical team were excluded from analyses.

Further patient information was obtained from the ED and inpatient electronic medical records. Information included patient demographics, ED diagnosis, Emergency Severity Index, disposition from the ED, admission location (if admitted), date and time of blood culture, unique accession number of the culture, site of culture obtained, organisms grown in the culture, if any antibiotic was given in the ED, route of administration of the antibiotic, time of administration of the antibiotic, and LOS.

We merged data from the laboratory database and the electronic medical record using the unique accession number for each blood culture and further confirmed the match using patient name, medical record number, and date and time of collection.

Charts of patients with contaminated blood cultures were reviewed to obtain information regarding the impact of the contaminated blood culture, and this information was used to provide individual nursing feedback.

Data were collected and stored into Research Electronic Data Capture, a secure Web-based data capture application.¹¹ 

The primary outcome measure was the percentage of contaminated peripheral blood cultures each month.

Our process measure was to ensure that nursing staff were following the change in collection process. We requested that nurses complete audit cards during each blood culture collection. The nurses were instructed to indicate in the audit card all the steps for blood culture collection they followed. Specifically, we measured the percentage of nurses who completed 75% to 100% of culture collection steps, as evidenced by documentation in the audit card. We also measured overall compliance with completing the audit cards.

There were several balancing measures. A primary concern from nursing staff before the start of the project was that the sterile collection process would lengthen the time to obtain a blood culture and, thus, delay the administration of antibiotics in sick patients. One balancing measure was the time interval between ordering an IV antibiotic to its administration to determine if this was prolonged after implementation of the QI initiative. We also measured ED LOS to determine if the process change increased time in the ED. We followed the percentage of peripheral blood cultures obtained relative to the total number of ED visits each month to ensure that the changes in contaminated blood cultures were not due to an overall reduction in blood cultures obtained in the ED. The QI initiative was implemented only in the ED. We compared the percentage of contaminated peripheral blood cultures in the inpatient units during baseline and intervention periods to determine if the reduction in contaminated blood cultures in the ED was due to the QI initiative versus an overall decrease in contaminated peripheral blood cultures in the hospital. Finally, we measured overall nursing acceptance of the process through a question on the audit card assessing whether they liked the new process or not.

Primary outcome and process measures were plotted over time in a statistical control P-chart by using established rules for identifying special-cause variation.^12,13  The balancing measure of time antibiotic was ordered to time given was plotted in an X-bar and S-chart. Characteristics of patients evaluated in the ED during baseline and intervention periods were compared. Categorical variables were analyzed by χ² and Fisher’s exact test and continuous variables with the Mann–Whitney U test. Data were analyzed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). A P value <.05 was considered significant.

Our university institutional review board determined that this QI initiative was exempt from human subjects’ review.

There were 1052 and 2204 peripheral blood cultures obtained during baseline and QI intervention periods, respectively. There were 71 (6.7%) contaminated peripheral blood cultures during baseline and 46 (2.1%) during the QI intervention. There were 136 commensals identified from 117 contaminated cultures. The most common pathogen was coagulase-negative staphylococci (59%), followed by viridans group Streptococcus (18%) and Staphylococcus epidermidis (7%) (Supplemental Table 2).

There was a threefold decrease in the percentage of contaminated peripheral blood cultures during the QI initiative (Fig 2). Analysis of the change over time revealed that the percentage of contaminated cultures started declining soon after the educational survey and continued to decline with implementation of further interventions.

There was no statistical difference between baseline and intervention groups in regard to sex, Emergency Severity Index, or ED discharge diagnosis (Table 1). There was a statistically significant difference in age, ethnicity, and admit department.

An audit card was completed for 36% of blood cultures collected during the QI initiative. Forty-four percent of audit cards were completed by a nurse auditing the procedure and 16% stated an ultrasound was used. Of those cultures with an audit card completed, 98% of nurses reported completing 75% to 100% of the steps of the new sterile process (Fig 3). The percentage of nurses who stated they liked or were neutral about the kit increased from 54% early in implementation to 93% by the end of the QI initiative (Supplemental Fig 6).

The average duration from the time an IV antibiotic was ordered to the time of administration was 55 minutes, and there was no significant change between the baseline and intervention periods (Fig 4).

There was no significant change in ED LOS for patients during baseline and the QI initiative (Table 1).

There was a small decline in the proportion of ED patients who received a peripheral blood culture during the baseline and intervention periods (3.8% vs 3.5%; P = .6). (Supplemental Fig 7). Our ED census declined in April 2020 secondary to the coronavirus pandemic.

There was no significant difference in the percentages of contaminated peripheral blood cultures on the inpatient units during the baseline and intervention periods (3.3% vs 2.9%; P = .65).

With our QI initiative, we successfully decreased the percentage of contaminated peripheral blood cultures threefold in our ED using a sterile blood culture collection process, nursing education, and individualized feedback. A retrospective case control study in our ED revealed a significant increase in repeat blood cultures (43% vs 1%), cerebrospinal fluid studies (10% vs 0%), and antibiotic administration (27% vs 1%) in children with contaminated peripheral blood cultures evaluated and discharged from the ED compared with controls.³  Reducing contaminated blood cultures in our ED thus reduced unnecessary painful procedures in children from additional bloodwork, IV starts, and lumbar punctures. Reducing contaminated blood cultures also reduces unnecessary use of antibiotics in children, which can lead to antibiotic resistance, as well as alteration in gut microbiome.¹⁴ 

Our previous study revealed that 52% of ED patients (evaluated and discharged from the ED) with contaminated peripheral blood cultures returned to the ED and/or were readmitted within 5 days of the initial visit because of contaminated cultures.³  This can pose a significant financial and emotional burden on families because of missed days from work, as well as child care for siblings while the parent is in the hospital. High financial distress and medical financial burden occur in families of hospitalized children of all financial brackets.¹⁵  Reducing contaminated cultures through our initiative can decrease this burden on families.

Our QI initiative used a multipronged strategy to reduce peripheral blood culture contamination. It has been revealed that implementation is more successful when people support a new strategy and value its outcomes.¹⁶  We achieved this first by educating nursing staff about the impact of contaminated blood cultures on children and families during the initial educational survey and video. We then ensured they supported the new strategy through feedback while rounding in the ED, at nursing staff meetings, and comments on the audit cards. The awareness of the consequences of contaminated cultures on children seen in our ED reinforced the risks of continuing with the status quo. This likely accounted for the initial reduction in contaminated cultures after the survey even before the change in the collection process.

Buy-in from nurses is easier when they feel empowered to make changes in processes they are involved in and feel that their feedback is translated into changes as well.¹⁷  Nurses provided input on the components of the sterile collection kit, and changes were made on the basis of their input while we trialed the kits in the ED, as well as at later stages of the project. Audit cards were used not only to determine compliance with the collection process but also to obtain ongoing feedback from the nurses.

El Faghaly et al¹⁸  used nursing education, standardized blood culture collection, and optimized blood volume to reduce rates of contaminated blood cultures from 2.85% to 1.54% in a tertiary care children’s hospital. However, we had been unsuccessful in reducing blood culture contamination in our ED despite multiple educational efforts, standardization of the collection process, and auditing of the process. In systematic reviews, educational strategies have been revealed to have only short-term effects for process change.¹⁹ 

In addition to educational strategies, we implemented a process change that we felt would be less susceptible to variations in nursing practices, nursing turnover, and decay with time. Hall et al⁴  used a sterile blood culture collection process to reduce rates of contaminated blood cultures from 3.9% to 1.6% in a tertiary care children’s hospital. Mullen et al¹⁰  used a venipuncture sterility checklist, phlebotomist feedback on contaminated cultures, and a blood culture ordering guideline to reduce contaminated blood cultures from 3.02% to 1.17%. Self et al⁹  used a sterile blood culture collection process, including a sterile kit to reduce contaminated blood cultures from 4.3% to 1.7% in an adult ED. We chose a similar sterile process change, along with continued educational efforts, including educational videos as well as audit and feedback techniques, including feedback on contaminated cultures, to successfully reduce our contaminated blood culture rates. We were able to validate that a sterile collection process is a viable option to reduce contaminated cultures even in a setting with a much higher baseline blood culture contamination rate than has been published in previous studies. A strength of our study was that we also had a temporal control (lack of change in contaminated cultures in inpatient units) that helped us demonstrate that the reduction in contaminated cultures in the ED was due to the sterile collection process.

There was a small decline in proportion of ED patients who received a peripheral blood culture after implementation of sterile collection (3.8% vs 3.5%). Our QI project may have increased awareness among providers of the effects of contaminated cultures, causing a decrease in ordering cultures that they may think would be uninformative, which we are exploring in a follow-up study.

A study by Hall et el⁴  revealed that each contaminated peripheral blood culture resulted in $2800 excess hospital charges. Decreasing contaminated blood cultures from 6.7% at baseline to 2.1% during the QI initiative potentially prevented ∼102 contaminated cultures, which would have incurred an additional $285 600. Our blood culture collection kits cost $10.71, with a total cost of $23 605 during the initiative. We had a net, estimated savings of ∼$261 995 because of the reduction in contaminated blood cultures. This does not account for the time saved by physicians to call families about contaminated blood cultures or the additional costs incurred by families in lost days at work due to follow-up visits at the primary care physician office, return to the ED, or admission to the hospital for contaminated blood cultures.

Limitations of this study include that our QI initiative was conducted in a single tertiary care children’s ED and may not be easily implemented at other hospitals. In addition, our laboratory personnel were invested in this initiative and were willing to assemble the kits, which may not be feasible at other hospitals. However, the kit consists of normal supplies to place an IV and obtain a blood culture, sterile towels, and sterile gloves. All these are readily available at other EDs and can be easily used to change the blood collection process to a sterile procedure. Finally, only 36% of audit cards were completed. Although it is possible those returned had a higher rate of compliance compared with those that did not complete the audit cards, the successful reduction in contaminated cultures suggests that majority of nurses did adhere to the sterile collection process.

Use of a sterile blood culture collection process, in addition to nursing education and individualized feedback, is an effective strategy to decrease peripheral blood culture contamination rates in a pediatric ED. The decrease in contamination rates benefit patient care through elimination of painful procedures, antibiotic use, and unnecessary hospitalization while also reducing health care costs.

We thank Stacy Herndon, RN, Angela Lee, MD, and Lauren Littell, MD for their contributions.